On weighted compactness of oscillation and variation of commutators associated with Schrödinger operators (Q6081110)

Given a family \(\mathcal{T}=\{T_\epsilon\}_{\epsilon>0}\) of integral operators \((T_\epsilon f)(x)=\int_{\vert x-y\vert >\epsilon} K(x,y) f(y)\, dy\) one defines the \(\rho\)-variation \[\mathcal{V}_\rho(\mathcal{T} f)(x)=\sup_{\epsilon_i\downarrow 0}\left(\sum_{i=1}^\infty \left\vert T_{\epsilon_{i+1}}f(x)-T_{\epsilon_i} f(x)\right\vert ^\rho \right)^{1/\rho}\] and the oscillation \[\mathcal{O}(\mathcal{T} f)(x)=\left(\sum_{i=1}^\infty \sup_{ t_{i+1}\leq \epsilon_{i+1}\leq t_i} \vert T_{\epsilon_{i+1}}f(x)-T_{\epsilon_i} f(x)\vert ^2 \right)^{1/2}\] where \(t_i\downarrow 0\).\par Given a Schrödinger operator \(\mathcal{L}\) for which \(\Gamma(x,y,\tau)\) is a fundamental solution for \(\mathcal{L}+i\tau\), for \(\ell=1,\dots, n\), one defines \[\mathcal{R}_{\mathcal{L}}^\ell (x,y)=-\frac{1}{2\pi}\int_{-\infty}^\infty (-i\tau)^{-1/2}\frac{\partial}{\partial x_\ell}\Gamma(x,y,\tau)\, d\tau\] and one defines the \(\ell\)th Riesz transform \[\mathbb{R}_{\mathcal{L}}^\ell f(x)=\lim_{\epsilon\to 0^+}\int_{\vert x-y\vert >\epsilon}\mathcal{R}_{\mathcal{L}}^\ell (x,y) f(y)\, dy\, .\] Schrödinger operators considered here have forms \(\nabla(-\Delta+V)^{-1}\nabla\), \(\nabla(-\Delta+V)^{-1/2}\), and other related forms where \(V\in B_{n/2}\). Here, \(B_q\) is the Hölder class of functions \(V\geq 0\) satisfying \((V^q)_B^{1/q}\leq CV_B\) for all balls \(B\subset\mathbb{R}^n\), where \(V_B=\frac{1}{\vert B\vert} V(B)\) with \(V(B)=\int_B V\).\par One defines a BMO class by first defining a critical radius \[\gamma(x)=\sup_{r>0}\left\{\frac{1}{r^{n-2}}\int_{B(x,r)} V(y)\, dy\leq 1\right\}\, .\] Set \[\Vert b\Vert_{\mathrm{BMO}_{\theta}(\gamma)} =\sup_{x\in\mathbb{R}^n, r>0}\frac{1}{(1+\frac{r}{\gamma(x)})^\theta\vert B(x,r)\vert}\int_{B(x,r)}\vert b(y)-b_B\vert \, dy\] and denote by \(\mathrm{CMO}(\gamma)\) the closure of \(C_c^\infty\) in \(\mathrm{BMO}(\gamma)\).\par One defines the commutator of \(b\) and \(\mathbb{R}_{\mathcal{L}}^\ell \) by \[\mathbb{R}_{b,\ell}^{\mathcal{L}}(f)(x)=\lim_{\epsilon\to 0^+}\int_{\vert x-y\vert >\epsilon}(b(x)-b(y))\mathcal{R}_{\mathcal{L}}^\ell (x,y) f(y)\, dy \, .\] Its adjoint \(\mathbb{R}_{b,\ell}^{\mathcal{L},\ast}\) is defined similarly. It is explained that these commutators are Calderón-Zygmund operators. Specific CZO estimates are provided. \par Finally, a maximal commutator of the form \[T_b^\ast f(x)=\sup_{\epsilon>0} \left\vert \int_{\vert x-y\vert >\epsilon} (b(x)-b(y)) K(x,y) f(y)\, dy\right\vert \] is discussed.\par The weight class \(A_p^{\gamma,\theta}\) consists of those weights \(\omega\geq 0\) for which there is a constant independent of \(B=B(x,r)\) such that \[\omega_B^{1/p} (\omega^{-p'/p})_B^{1/p'}\leq C\left(1+\frac{r}{\gamma(x)}\right)^\theta\, .\] The main results are Thms.~3.1, 4.1 and 5.1. \par Theorem 3.1 states that if \(\rho>2\), \(\theta>0\) and \(V\in B_q\) (\(q>n/2\)), \(p\in (1,\infty)\), \(b\in \mathrm{CMO}_\theta(\gamma)\) and \(\omega\in A_p^{\gamma,\infty}\), then the variation operators \(\mathcal{V}_\rho(T_{b,\epsilon})\) are compact from \(L^p(\omega)\) to itself. Under the same hypotheses, Thm.~4.1 states that the oscillation operators \(\mathcal{O}(T_{b,\epsilon})\) are compact from \(L^p(\omega)\) to itself. Theorem 5.1 establishes corresponding compactness properties of the variation operators \(\mathcal{V}_\rho(\mathbb{R}_{b,\ell,\epsilon}^{\mathcal{L}})\) and \(\mathcal{V}_\rho(\mathbb{R}_{b,\ell,\epsilon}^{\mathcal{L},\ast})\) and oscillation operators \(\mathcal{O}(\mathbb{R}_{b,\ell,\epsilon}^{\mathcal{L}})\) and \(\mathcal{O}(\mathbb{R}_{b,\ell,\epsilon}^{\mathcal{L},\ast})\), under suitable conditions on \(\mathcal{L}\), and on the index of the class \(B_q\) defining \(b\). The results are placed in historical context in the introduction.